(A) Example traces of MA currents in WT and Elkin1-KO clones of WM266-4 cells on LM511-coated pillar arrays. A residual current is present in Elkin1-KO clones at large deflections. (B) Stimulus-response plots showing that MA currents are significantly inhibited in the Elkin1-KO clones in comparison with the WT clones (ordinary two-way ANOVA, Elkin1 KO = 19 cells, WT = 11 cells, ****p<0.0001; Sidak’s multiple comparison, **p=0.007, *p=0.03, **p=0.004). (Data displayed as mean ± s.e.m.). (C) A transwell analysis of migration onto LM511-coated membranes shows that WT clones (3B6 and 3E9) were indistinguishable from WT controls, whereas Elkin1-KO clones (3C6 and 3D6) exhibited significantly reduced transmigration (one-way ANOVA, parental control n = 8 wells, 3E9 = 4 wells, 3B6 = 7 wells, 3D6 = 8 wells, 3C6 = 8 wells, ****p<0.0001; Dunnett’s multiple comparisons, control vs 3C6, ****p<0.0001; control vs 3D6, ****p<0.0001, samples normalised against WT control). (D) The transmigration phenotype in the KO was rescued by overexpression of hsElkin1-iso3, but not hsElkin1-iso3-L210F (one-way ANOVA, KO control n = 11 wells, +hsElkin1-iso3 = 12 wells, +hsElkin1-iso3-L210F = 11 wells, ****p<0.0001; Dunn’s multiple comparisons, Control vs +hsElkin1-iso3, ***p=0.0006; +hsElkin1-iso3 vs +hsElkin1-iso3-L210F, ***p=0.0002, samples normalised against KO control). (E) In the A375 melanoma cell line, an Elkin1-KO clone also exhibited a significant decrease in transmigration onto LM511, in comparison with a WT control (unpaired t-test with Welch’s correction, WT and Elkin1-KO = 8 wells, **p=0.002, samples normalised against WT control). (C–E) Individual data points overlay mean ± s.e.m. See Figure 5—figure supplement 1 for CRISPR/Cas9 strategy and knockout clone validation.